Candidates About

Taixing Bi

OBJECTIVE

§  A bright, talented and self-motivated data scientist/machine learning engineer seeking a position to draw insights from a variety of large data sets using Python wth Sci-kit, R, SparkML, Java and SQL.

RELEVANT EXPERIENCE

Data Scientist                                                                                                                                                                            CGMAX

New York City, New York                                                                                                                                       June.2017 –

Customer segmentation with Python Sklearn (dimensional reduction and clustering)

·         Feature selection, relevance, visualize feature distributions and outlier detection(Tukey‘s method) with Pandas.

·         Feature transformation: princple componet anaylys(PCA) and visualized with Seaborn and Pillow.

·         Segmentation: Clustered samples with K-mean and visualized using python with flask.

Customer classfication based on CNN with Tensorflow (deep learning)                                                              

·         Selected feature and labeld sample from Hadoop.

·         Trained, hyper-paremeters tuned and tested model(60%, 20%, 20%) with virtual samples, then fine tuning with real samples and final tested in real environment based on GPU(Titan).

·         Achieved 91% accuracy.                  

Product sales prediction with LSTM model (regression model)                                                                         

·         Collect data and filter high leveage point since 2000 and predicted production volume for next month.

·         Set up LSTM of RNN model using Python3, Keras with Tensorflow, fine-tune parameteres: neurons, epochs and decay, and visualization.

·         Error accuracy is low to 0.00029 MSE (0.02 RMSE) 

Machine Learning Engineer                                                                                                                                                  NVIDIA

Santa Clara, CA                                                                                                                                  Jan.2016 – June.2017

Customer review classification (Natural Language Processing of Document Classification)                           

·         Cleaned, tokenized and embedded document(glove.6B.100d) with Keras.

·         Trained CNN with shallow model with  paremeters tuning.

·         Accuracy is 71.9%.

Credit Card Fraud Detection with Autoencoder model (hidden neural network)                                                 

·         Visualized transaction class distribution for highly imbalanced dataset using Matplotlib.

·         Trained autoencoder model with epoch 100, batch 32 using Keras with Tensorflow training on Linux.

·         MSE analysis and  ROC using Pandas,  AUC is 0.9618.             

Teaching Assistant                                                                                                                     University of Alabama in Huntsville

Huntsville, Alabama                                                                                                                             Feb. 2015 – May. 2015

Python Introduction                                                                                                          

Data Scientist                                                                                                                                                                         XJ Finance

China                                                                                                                                                                April.2010 – April.2013

Credit Risk with SparkR (Decision Tree)

·         Collected and selected feature loan status, amount, interest rate, grade, home ownership, income, age, etc..

·         Filtered outlier and visualized feature with Tableau.

Accuracy of Decision Tree and logistic regression is 88.8% and 74.2%. ORC is 0.5997 and 0.658.               

EDUCATION

·         University of Alabama in Huntsville, Huntsville, AL                                                   Graduated Jan.2016

Master of Science in Electrical and Computer Engineering

Technical Skills                                                                                                        

Computer Software: Python, C/C++, R, Hadoop, Spark, Matlab, SQL and Tensorflow, Keras.

Language: English and Chinese.